Microsoft Word - 2026-06-26 Boundy letter re 0651-0089 DOCX ICR 260626 2124 DEB

Status: Original and derived artifacts are available for this document.

Download: pdf

David Boundy
P.O. Box 590638
Newton, Mass. 02459
646.472.9737
[email protected]
www.linkedin.com/profile/view?id=482230

June 26, 2026
Via regulations.gov
Justin Isaac and Raul Tamayo
Office of the Chief Administrative Officer
U.S. Patent and Trademark Office
P.O. Box 1450
Alexandria VA 22313-1450
Dear Mr. Isaac and Mr. Tamayo:
The USPTO should amend this ICR to request continued approval for permitted DOCX
filing, but to remove any request for clearance for the § 1.16(u) surcharge for non-DOCX filing.
As even the USPTO admits (by failure to explain to the contrary), DOCX cannot possibly and
does not work reliably. Users should be permitted to file in whatever full-text form their word
processor reliably outputs, whether that be DOCX, PDF, or PDF/A, at the user’s choice, with no
surcharge.
At the time the USPTO promulgated this regulation, the PTO estimated its cost savings
per filing at $3.15 per application.1 Now, even on the laughably-low estimate of 0.5 hours that
the USPTO finally concedes (now abandoning its 2019-2023 estimates of zero), the cost to the
public is $223.50 per application. A 71-to-1 ratio is “shifting disproportionate costs or burdens
onto the public” forbidden by 5 C.F.R. § 1320.5(d)(1)(iii). 5 U.S.C. § 1320.5(d)(1)(i) requires the
agency to “demonstrate that it has taken every reasonable step to ensure that the proposed
collection of information … Is the least burdensome necessary for the proper performance of the
agency‘s functions.” By the agency’s own admission, PDF filing is “the least burdensome”
approach.
Further, as noted in § 2 of this letter, the USPTO’s burden estimates are laughably low,
by a factor of 10.
And finally, 44 U.S.C. § 3506(c)(3)(B) and 5 C.F.R. § 1320.9(b) require the USPTO to
limit collecting information that is “unnecessarily duplicative.” 5 U.S.C. § 1320.5(d)(1)(ii)

1

U.S. Patent and Trademark Office, Setting and Adjusting Patent Fees During Fiscal Year 2020, 85
Fed. Reg. 46932, 46947 col. 2 (Aug. 3, 2020).

DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459 | T: +1 646.472.9737

David Boundy

Page 2 of 17

U.S. Patent and Trademark Office
June 26, 2026
elaborates: the agency “shall demonstrate that it has taken every reasonable step to ensure that
the proposed collection of information … Is not duplicative of information otherwise accessible
to the agency.” the USPTO itself—and every known private sector guide—recommends as best
practice that DOCX filing be accompanied by an auxiliary PDF. The USPTO has never offered
a good explanation for how a $3.15 per application savings can justify a duplicative collection.
The underlying factual dispute is that the USPTO claims that DOCX is designed as an
interoperable format that guarantees that content will be rendered consistently on all computers,
and that that consistency will be preserved over time with new software releases. The DOCX
standards documents make no such representation, and make a number of contrary
statements, discussed in the table at page 8 of this letter.2 Microsoft’s documentation state to
the contrary (see discussion of “extensions” at pages 8 and 10). Many members of the public
have written to provide examples of where a single DOCX file appears differently when viewed
or printed from different computers. Particularly, mathematical formulas change remarkably
when a DOCX file is generated on one computer and rendered on another. Even base text can
change. The USPTO has never confronted the contrary evidence, except to make unsupported
claims that are contrary to all evidence, and that demonstrate lack of understanding of basic
technological vocabulary.
In contrast, PDF, especially the PDF/A variant (“A” is for “archival”) is designed
specifically and exactly for the purposes the USPTO desires (and that’s why it’s the format
used by federal courts and all other agencies. The PDF/A standard3 states that PDF/A is
designed to “maintain[ ] … properties, such as the logical organization of pages,
sections, and paragraphs, machine recoverable text stream in natural reading order.”
There is no analogous commitment in the DOCX standard. The Microsoft documentation says
the opposite.4 The USPTO has never identified any reason to believe the standards document.
In its 2023 ICR Supporting Statement,5 the Office claimed that “In comparison to DOCX,
rich text formatted PDF does not provide a reliable source of structured text.” In spite of the
obligation of 44 U.S.C. § 3508(c)(3) and 5 C.F.R. § 1320.8(a)(4) for “objective support,” the
USPTO stands on nothing more than personal opinion of Mr. Isaac and Mr. Tamayo. The
USPTO has produced no “objective support” for this claim. The only document generated by

2

ECMA-376, Office Open XML file formats, https://ecma-international.org/publications-andstandards/standards/ecma-376/ and ISO/IEC 29500-1:2016, Information technology — Document
description and processing languages — Office Open XML File Formats,
https://www.iso.org/standard/71691.html ECMA-376 is available for free at the preceding URL. I
uploaded the relevant parts of ISO/IEC 29500 at the OMB web site,
https://www.reginfo.gov/public/do/PRAViewDocument?ref_nbr=202309-0651-002 There’s really no
excuse for not reading it.
3

ISO 19005-4:2020 Document management — Electronic document file format for long-term
preservation, Introduction, page vi, https://www.iso.org/standard/71832.html or https://www.pdfainc.org/cart/?add-to-cart=594
4

Microsoft, [MS-DOCX]: Word Extensions to the Office Open XML (.docx) File Format,
https://learn.microsoft.com/en-us/openspecs/office_standards/ms-docx/b839fe1f-e1ca-4fa6-8c265954d0abbccd

5

Supporting Statement,
https://www.reginfo.gov/public/do/DownloadDocument?objectID=148432201 (Dec. 7, 2023).
DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM

David Boundy

Page 3 of 17

U.S. Patent and Trademark Office
June 26, 2026

the USPTO is the “yearlong study” referred to in the Final Rule notice.6 That “yearlong study”
was withheld from the public during rulemaking (unlawfully). When we obtained it by FOIA, it
stated the opposite of the USPTO’s representation now: that PDF was “the right approach.”
The USPTO has produced no document that suggests that DOCX was designed uniformity as a
design consideration; the DOCX standards documents and Microsoft’s own documentation state
otherwise. The USPTO has offered no counter to the PDF/A standard document that PDF/A is
designed to “maintain[ ] … properties, such as the logical organization of pages, sections, and
paragraphs, machine recoverable text stream in natural reading order.”7 It’s clear that the
USPTO has never actually looked at the standard document: the URL in the Final Rule notice
and 2023 Supporting Statement for ISO/IEC 29500 is wrong.8 And the USPTO has never given
any meaningful response to the public’s comments advancing PDF/A as the solution to the
USPTO’s problem.
Perhaps, the USPTO’s claims that DOCX is a “reliable source of structured text” and
PDF/A is not can be excused as aggressive advocacy of technologically ignorant USPTO
employees who are unfamiliar with basic technological vocabulary, and ignorant of the
technology of document preservation. Below, we present evidence that DOCX is not reliable,
was not designed to be reliable, and PDF/A is. Unless the USPTO engages issue by issue with
“objective support,” any continued representation that DOCX is “a reliable source of structured
text” may be taken as intentionally false.

1.

A recent personal observation that DOCX files appear differently
even on two different versions of Word

Many letters from the public have made the same point: the DOCX standard does not
guarantee interoperability or reproducibility across implementations or across time.
In February 2026, I had a direct example. A client sent me a draft patent application that
he had composed in Microsoft Word with the help of AI. I opened it in Word for Microsoft Office
Professional Plus 2021. I saw this, obviously incorrect:

I went to a different computer, one that runs Microsoft 365, the Software-as-a-Service version. I
opened the same document (or rather a different save out of the same email, so the identical
document content but a different copy) and opened it and got this:

I don’t know whether the problem is that the two different versions of Word, or different fonts
loaded on two different computers, or different equation editor plug-ins. But what is clear is that

6

An entire copy of the “yearlong study” obtained by FOIA is provided as Exhibit A in the Letter 152
Patent Practitioners, https://www.reginfo.gov/public/do/DownloadDocument?objectID=135639300

7

See note 3 supra.

8

It is apparent that the PTO has never read the standard hasn’t a clue whist it’s talking about. The
URLs given in the PTO’s Federal Register notices and OMB Supporting Statements are incorrect. See
footnote 2 supra.
DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM

David Boundy

Page 4 of 17

U.S. Patent and Trademark Office
June 26, 2026
neither the ISO 29500 standard nor Microsoft guarantee that a given DOCX file will open the
same on two different computers. In fact, Microsoft all but promises the opposite.9
Microsoft does not even guarantee that a DOCX will open the same across two different
installations of Word.4 There is no possible way that a third-party product (like the USPTO’s
DOCX intake) can guarantee consistency, if DOCX format changes over time. Since the
USPTO does not tell us which temporal dialect of DOCX they implement (and they can’t
possibly keep up with Microsoft’s changes, which occur several times a year4 and are reflected
immediately in Office 360 SaaS, with no apparent warning to third-party vendors), the USPTO
cannot possibly guarantee correctness over time.

2.

The 2026 Notice Omits significant burden

The April 2026 60-day notice gives the following burden estimates. The notice gives no
rationale or basis for the “estimated time.”

(A)

The USPTO’s estimates are low by factor of ten, underestimated by
excluding proofreading burden

The USPTO’s 60-day notice gives no basis whatsoever for the estimates of 0.5 hours;
the number appears to be pulled out of the thinnest of thin air.
0.5 hours squares with the cost of managing three duplicative copies ((a) the usergenerated PDF copy that was generated and was the upload document before 2023 and that is
still, by best practice above, generated and submitted as the “auxiliary PDF” check document,
(b) the DOCX copy originally generated by the user and uploaded, and the (c) “feedback
document” DOCX generated by the USPTO, which must be saved and compared to (b)) To all
appearances, the USPTO has booked only that 0.5 hours.
However, burden includes at least two more components: “completing and reviewing” or
proofreading, 44 U.S.C. § 3502(2)(E), and filing correction papers.

9

Word Extensions of OOXML, note 4 supra.
DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM

David Boundy

Page 5 of 17

U.S. Patent and Trademark Office
June 26, 2026
In the Federal Register, the USPTO acknowledged that the USPTO’s processing of
DOCX, was error prone, and “strongly encouraged” proofreading:
Applicants are strongly encouraged to review their applications, including the
USPTO-generated PDF, shortly after filing the application to identify any errors or
discrepancies in the record, as discussed above. The applicant should file any
necessary petition to correct the record early in prosecution and promptly after
discovering any errors or discrepancies.10
Similarly, a recent post11 by Pushpendra Dwivedi on IPWatchdog.com (one of the three
most-viewed blogs in the patent field) reminded the public that DOCX filing is more risky than
PDF filing, and that protecting against those risks requires certain “best practices.” This is a
direct quote, except to replace a bullet list with numbered items so that our comments can refer
to those numbers:
[T]he proposed DOCX filing system has raised concerns among practitioners. They fear that the
USPTO’s conversion of the “User-Submitted DOCX” into a “USPTO-generated DOCX” and
“USPTO-generated PDF” could alter the application content or lose evidence, since the “Usersubmitted DOCX” is not preserved. Moreover, the editable nature of DOCX could lead to
unintentional changes before filing.
Mastering the Art of DOCX Submissions: Best Practices
Now that the fundamental understanding of DOCX filing is established, let’s delve into mastering it.
Consider adopting these best practices when submitting your documents in DOCX format:

1. Follow the USPTO guidelines when you prepare your application in DOCX format to
avoid errors during the filing process.
2. Save the feedback document that Patent Center creates when you submit your
application. This document may serve as evidence in case of any litigation related to
misuse.
3. Submit a supplementary PDF along with the DOCX filing by 17th January 2024, unless
the PTO extends the deadline.
4. Review your submission a month or two after filing and before the 1-year filing deadline
expires to fix any issues that you may have overlooked initially.
5. File your documents early, so that you and your staff or attorneys can examine the
feedback report and take the necessary actions.
6. Compare the documents to ensure no unintentional changes are there. Mark this is a
mandatory step. You must compare the User-submitted DOCX to the USPTO-generated
DOCX; the User-submitted DOCX to the USPTO-generated PDF; a User-submitted PDF
to the USPTO-generated DOCX; and the User-submitted PDF to the USPTO-generated
PDF.

10

USPTO, Filing Patent Applications in DOCX Format, 87 FR 25226, 25228 (Apr. 28, 2022)

11

Pushpendra Dwivedi, Mastering USPTO DOCX Formats: The Ultimate Guide,
https://ipwatchdog.com/2024/03/11/mastering-uspto-docx-formats-ultimate-guide (Mar. 11, 2024)
DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM

David Boundy

Page 6 of 17

U.S. Patent and Trademark Office
June 26, 2026
One comment to this post reminded
Just realize that 99-plus percent of the time it works…If 99% of my deposits in a bank were fine, I
would still be very upset about the 1% where they got it wrong.
Several other blog posts note the importance of careful proofreading to overcome the
error-prone DOCX filing;12

(B)

The USPTO’s estimates to not include plausible estimates of error
correction filings

In past filings, the USPTO acknowledged that some number of requests for correction
had to be filed. The USPTO failed to book that time in 2023, and now books these “best
practices” at zero.
Past letters to the USPTO and to OMB have estimated the burden of proofreading (Mr.
Dwivedi’s bullets 4 and 6) plus error correction. These past letters noted that a small number of
these proofreading exercises will locate an error, and that correcting that error will be very
expensive. Nonetheless, careful proofreading is required, because when the USPTO
introduces an error, there may be no opportunity whatsoever to correct it (if it forms the basis for
a foreign filing), and the cost of seeking correction will be disproportionately high. The estimates
in the 60-day notice ignore these large cost drivers—they may be few in number, but they must
nonetheless be included in the average, which drives the total to $200 million to $600
million.13,14,15,16

12

PatentDocs, USPTO Allows Applicants to File PDF with DOCX Submissions on Temporary Basis,
https://patentdocs.org/2022/05/03/uspto-allows-applicants-to-file-pdf-with-docx-submissions-ontemporary-basis; Ant-Like Persistence, USPTO fails to support DOCX from non-Microsoft word
processors,, https://blog.oppedahl.com/uspto-fails-to-support-docx-from-non-microsoft-wordprocessors/
13

Estimate of Kilpatrick Towensend, estimates proofreading at 3 hours, and offers no estimate of the
cost of correction. https://www.reginfo.gov/public/do/DownloadDocument?objectID=135640600

14

Estimate of David Boundy, Exhibit 7 to letter of 152 Patent Practitioners, note 6 supra, at Exhibit 7,
pages 3-5.

15

Bradley Forrest, a former Group Chief IP Counsel at IBM estimates that the proofreading costs
that must be invested in every patent application, and the error-correction costs that must be borne
rarely but at great expense, total at least $200 million per year. Estimate of Bradley Forrest, Exhibit 6
to letter of 152 Patent Practitioners, note 6 supra.
16

Consensus estimate of One Hundred Seven Patent Practitioners is about $600 million.
https://www.reginfo.gov/public/do/DownloadDocument?objectID=135640900
DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM

David Boundy

Page 7 of 17

U.S. Patent and Trademark Office
June 26, 2026

3.

The USPTO’s DOCX rule (used under acknowledged best
practices) requires duplicative information collection

The USPTO’s “DOCX Filing Tips & Best Practices” recommends filing an auxiliary
PDF. The USPTO has submitted no evidence—only unsupported bloviation—that a PDF
submission (generated in text form, as almost all are) is in any way insufficient to meet all
requirements that the USPTO has (see the table row discussing “preservation” and “natural
reading order” of PDF/A, on page 9).
17

4.

DOCX does not purport to be an “interoperability” standard –
the USPTO’s past claims are false

In past Federal Register notices and regulatory filings at OIRA, the USPTO has claimed
that DOCX is “a reliable source of structured text”18 and “presents documents, including text
and formatting, in a manner that is independent of software, hardware, or operating systems.”19
The USPTO has cited no source for these claims—they appear to be made up out of thin air.
Not all standards purport to achieve the same ends. Some standards merely purport to
set a standard for a minimum conforming implementation. For example, the Fortran, C, C++
and similar language standards define a minimum baseline for a conforming implementation,
but do not purport to allow interoperability. A legal program in any of these languages can yield
entirely different results on different computers. ISO 29500, the DOCX standard, is this kind of
non-interoperability standard.
In contrast, some standards are “interoperability” standards. An interoperability standard
ensures that all conforming implementations will exactly work as intended with all others.
Examples include cell phone standards, WiFi, PDF, FHIR (Fast Healthcare Interoperability
Resources), and FIX (Financial Information Exchange). PDF is an interoperability standard: the
first sentence on the ISO web page20 states “This document specifies a digital form for
representing electronic documents to enable users to exchange and view electronic documents
independent of the environment in which they were created or the environment in which
they are viewed or printed.”
Standards bodies such as ETSI21 and Oasis22 have written guidelines for interoperability
standards, and steps that must be taken for them (as distinguished from non-interoperability

17

See DOCX Filing Tips & Best Practices,
https://www.uspto.gov/sites/default/files/documents/DOCX_Filing_Tips_and_Best_Practices_Octobe
r_8_2024.pdf
18

Supporting Statement, note 5 supra, Response to Comment 3.

19

Supporting Statement, note 5 supra, Response to Comment 2.

20

ISO 32000-2:2020 Document management — Portable document format,
https://www.iso.org/standard/75839.html
21

ETSI, Solve the Challenge of Interoperability! Interoperability Best Practices,
https://www.etsi.org/images/files/Events/interoperability_best_practices_handbook.pdf

22

Oasis Open, Interoperability Guidelines, https://www.oasis-open.org/policiesguidelines/interoperability-guidelines Oasis explains the difference between interoperability
DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM

David Boundy

Page 8 of 17

U.S. Patent and Trademark Office
June 26, 2026
standard), an interoperability standard, that avoids vendor lock-in or ambiguous outcomes, must
rely on specific core principles.
•

Closed Feature Set (No Extensions): The specification must be normative and
bounded¸ and must not allow vendor-specific extensions or subsetting. Compliance
means supporting the entirety of the base standard and nothing else. An “extension” is
a feature that exists in one implementation but not in others; therefore files created with
the extension cannot be processed by other standard-conforming software. (Major
vendors use extensions to lock customers in; extensions are anathema to
“interoperability” standard.) “Subsetting” is when an implementation implements and
permits less than the entire standard. Subsetting, likewise, means that a file created on
one computer cannot necessarily be processed correctly on another.

•

Deterministic Output: The standard must define explicit and invariant requirements
around input-to-output mappings. The standard must permit no “implementation
defined” behaviors (the standard may allow implementation maximum limits, but must
guarantee consistent behavior within those limits).

•

Strict Conformance Testing: A robust, unambiguous suite of test assertions must
exist. This validates whether independent implementations conform perfectly to the
rules, confirming behavior is identical across the ecosystem.

•

Platform & Vendor Neutrality: The definitions must remain divorced from specific
internal system architectures, programming languages, or operating environments.

ISO 19005-4
The PDF/A-4 Standard23
(and ISO 32000 the base PDF standard)
PDF is the product of an industry-wide
“voluntary conformance standard” process.
The PDF Association has over 125 paid
members, including Adobe, dozens of
competing software vendors, the United States
Government Publishing Office, National
Archives and Records Administration, and
Library of Congress, analogous government
offices of India, Canada, the Netherlands,
Denmark, Australia, New Zealand, The UK
Association for Accessible Formats (UKAAF),
Boeing, Jet Propulsion Labs, and DocuSign.25

ISO/IEC 29500
The DOCX standard24
DOCX is a single-vendor standard.26 The
ECMA-376 and ISO committees were stacked
by Microsoft (for example, both chairs were
Microsoft employees), and the standardization
process was fast-tracked by compressing the
process into a single 5-day meeting to
consider a 7000-page document, that
considered only 20% of the amendments
proposed by ECMA.27 Microsoft publicly
admitted that one of its representatives had
engaged in vote buying.28

standardization and non-interoperability standardization, and standards that are “tightly bound to the
needs of that one vendor.”
23

PDF/A standard, ISO 19005-4, note 3 supra, Introduction, page vi.

24

ISO/IEC 29500 OOXML, note 2 supra.

25

https://pdfa.org/member/
DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM

David Boundy

Page 9 of 17

U.S. Patent and Trademark Office
June 26, 2026
The first paragraph of the PDF/A-4 standard
states that the standard is directed to ensuring
that all future users of a PDF/A document
“maintaining their visual appearance as
well as their higher-order properties, such
as the logical organization of pages,
sections, and paragraphs, machine
recoverable text stream in natural reading
order.”29 The third paragraph reads “The
primary purpose of ISO 19005 is to define a
file format based on PDF, known as PDF/A,
which provides a mechanism for representing
electronic documents in a manner that
preserves their visual appearance over
time, independent of the tools and systems
used for creating, storing or rendering the
files. ”

The DOCX Standard makes no analogous
promise. The USPTO has asserted this
repeatedly since 2020, but has never identified
any document that makes such a promise, let
alone provided a cite or other “objective
support.” The USPTO’s assertion appears to
be nothing more than hopeful opinion.

26

The ISO/IEC 29500 and ECMA-376 standard (also known as Office Open XML or OOXML) is
widely considered in the industry to be essentially Microsoft’s proprietary file specification, that
Microsoft shepherded through the standardization process under its heavy influence. Microsoft
dominates this standard in several concrete ways. (a) Microsoft submitted the massive 6,000+ page
OOXML specification to ECMA International in 2006. The ECMA committee was chaired by two
Microsoft employees. The ECMA-376 standard—6000 pages—emerged in only a year from
submission to approval. Microsoft then pushed to make ECMA-376 an ISO/IEC 29500 standard,
and fast-tracked it in a few months. That proceeding was heavily contested, with critics, competitors,
and standards advocates alleging vote buying and manipulation, heated verbal confrontations,
refusal to come to consensus, and procedural irregularities. Detractors argued it was too complex
and tailored specifically to Microsoft's pre-existing software architecture. Wikipedia, Standardization
of Office Open XML, https://en.wikipedia.org/wiki/Standardization_of_Office_Open_XML; Network
World, Office Open XML standardization to drag into next year, note 28, infra; The Document
Foundation, Why OOXML is not a standard format for office documents,
https://blog.documentfoundation.org/blog/2026/02/06/why-ooxml-is-not-a-standard-format-for-officedocuments/
27

https://blog.documentfoundation.org/blog/2026/02/06/why-ooxml-is-not-a-standard-format-foroffice-documents; Jira Kosek, From the Office Document Format Battlefield, IEEE IT Professional
10:3:51-55 (May-Jun 2008). Articles explaining the irregularities include Microsoft Allegedly Bullies
and Bribes to Make Office an International Standard, https://www.wired.com/2007/08/microsoftallegedly-bullies-and-bribes-to-make-office-an-international-standard (Aug. 31, 2007); Office Open
XML ISO certification process grows even murkier for Microsoft, https://arstechnica.com/informationtechnology/2007/07/office-open-xml-iso-certification-process-grows-even-murkier-for-microsoft/ (Jul.
26, 2007); Complaints about the national bodies process,
https://en.wikipedia.org/wiki/Standardization_of_Office_Open_XML#Complaints_about_the_national
_bodies_process. The PTO cannot claim to be ignorant of this history—the PTO was given a copy of
this article which explains Microsoft’s sordid behavior. 60-day letter of 152 Practitioners
28

Network World, Office Open XML standardization to drag into next year,
https://www.networkworld.com/article/813817/software-office-open-xml-standardization-to-drag-intonext-year.html
29

See note 3 supra.
DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM

David Boundy

Page 10 of 17

U.S. Patent and Trademark Office
June 26, 2026
PDF/A does not permit extensions. (PDF
2.0 permits extensions under carefullycontrolled circumstances. The whole point of
PDF/A is to exclude any features of full PDF
that introduce dependencies outside the four
corners of the PDF file).
PDF/A does not permit subsetting, except
that an embedded font may be subsetted to
only those characters actually used.
The PDF/A standard document never uses the
word “implementation defined” or any
synonym. (The base PDF standard allows
“implementation defined” limits for maximum
sizes of internal tables, but every behavior is
specified; no behavior is left “implementation
defined.”
Test suites were developed by the PDF
Association and veraPDF consortium.33 A
vendor-neutral test suite is essential so that
every implementer of PDF has a way to
validate that its software conforms to the
standard.

PDF/A is a “live” standard, updated regularly

30

See note 4 supra.

31

ISO/IEC 29500 OOXML § 2.4 note 2 supra.

Microsoft has a 118-page document explaining
extensions that are added every month or
two.30 If a patent applicant uses one of these
features when newly added to Office 365,
there is no possible way that the USPTO’s
DOCX intake software can process it correctly.
ISO/IEC 29500 permits subsetting. The
application need not implement operations on
all XML elements defined in ISO/IEC 29500.”31
ISO/IEC 29500 has many dozens of
“implementation defined” behaviors, some of
which were pointed out in one of the comment
letters.32 “Implementation defined” and
“interoperable” are mutually exclusive
properties for a standard document.
A web page34 claims that an OOXML validator
exists, but then the page admits that the tool
doesn’t exist, never has, and is only “Tool
coming soon.” There is no indication that this
tool is to be developed in consultation with any
standards body. The “What kind of issues are
detected” line does not list the issues relevant
to reproducibility.
Microsoft has never committed to observing
the standard – Microsoft has a 118-page
document describing how Word departs from
the standard. https://learn.microsoft.com/enus/openspecs/office_standards/msdocx/b839fe1f-e1ca-4fa6-8c26-5954d0abbccd
The DOCX standards, ECMA-376 and
ISO/IEC 29500-1:2016, have not been
updated (at least not the relevant parts) since
initial adoption in 2016

32

Seventy-Three Patent Practitioners, Comment letter on Setting and Adjusting Patent Fees During
Fiscal Year 2020,
https://www.uspto.gov/sites/default/files/documents/Comment_Seventy_Three_Patent_Practitioners
_092719.pdf at pages 15-16 (Sep. 27, 2019)
33

https://pdfa.org/resource/isartor-test-suite and https://verapdf.org/ (specifically
https://pdfa.org/resource/verapdf-test-suite and https://verapdf.org/category/software )
34

OOXML Validator | AnyOnlineTool
DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM

David Boundy

Page 11 of 17

U.S. Patent and Trademark Office
June 26, 2026
The Library of Congress warns that conversion between LibreOffice’s ODT and Word’s DOCX is
neither straightforward nor reliable:35
Conversion between DOCX and ODT: Acknowledging the interest in whether
conversion between DOCX and ODT (Open Document Format word-processing)
files could be reliable, ISO started a work item to explore this issue. ISO/IEC TR
29166:2011 Information technology -- Document description and processing
languages -- Guidelines for translation between ISO/IEC 26300 and ISO/IEC
29500 document formats is the output of that expert working group. The report
documents the challenges of translation between OOXML and ODF formats,
including the word-processing formats, based on the standards as documented
at the time. This report, available from ITTF, describes features and functionality
for the three primary types of office document and characterizes the
translatability of features and functions as high, medium, or low. The challenges
are significant since the two formats use different underlying models. Although
simple documents can be effectively converted, a round-trip to an identical
document should never be expected. Display differences will be common after
conversion, most of no semantic significance, but many resulting in different
pagination or spacing.

5.

Specific comments to which commenter requests direct
answers, supported by evidence

Following are several specific questions that request answers. Each questions tests
facts and assumptions underlying whether the collection of information is necessary for the
proper performance of the functions of the Agency, accuracy of the Agency’s estimate of
burden, validity of the methodology and assumptions used; quality, utility, and clarity of the
information to be collected; and whether the agency has acted out of technological
misunderstanding that increases burden on those who are to respond.
Please respond fairly. Please do not recharacterize comments—either the comments
above or the questions below. Please do not respond with non sequiturs that evade the
comment. Please familiarize yourself with terms of art such as “interoperability standard” and
“open source,” so that your responses do not demonstrate the buffoonery of not knowing the
meanings of technological terms. Please do not demonstrate willful avoidance of learning what
words mean and their relevance. Under 44 U.S.C. § 3506(c)(3) and 5 C.F.R. § 1320.9, the
Office is required to “provide a record supporting” each of its claims—please cite evidence for
the USPTO’s positions, and please do not stand on personal opinions. Please do not contradict
statements in the three relevant standards (ISO/IEC 29500 (the DOCX standard), ISO 19005-4
(the PDF/A-4 standard), and ISO/IEC 32000 (the base PDF standard)) without explaining in
detail why the USPTO believes that the documents do not mean what they say. Please do not
attribute to standards documents promises that they do not state—for example, ISO/IEC 29500
makes no claim whatsoever that DOCX provides a reproducible, uniform, consistent “reliable
source of structured text" (in fact, by reserving "implementation defined" behaviors, it promises

35

Sustainability of Digital Formats: Planning for Library of Congress Collections, DOCX Transitional
(Office Open XML), ISO 29500:2008-2016, ECMA-376, Editions 1-5,
https://www.loc.gov/preservation/digital/formats/fdd/fdd000397.shtml
DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM

David Boundy

Page 12 of 17

U.S. Patent and Trademark Office
June 26, 2026
the opposite) -- if the USPTO believes otherwise, please identify and cite specific statements in
the standard document.
Citations to evidence are essential to honesty and accuracy. The DOCX standard,
ISO/IEC 29500, is not at https://www.iso.org/committee/45374/x/catalogue The USPTO has
given this cite repeatedly, from the NPRM in 2019 to the Supporting Statement in 2023. That
URL is wrong. This repetition of uncorrected error strongly suggests that the USPTO acts with
contempt for fact and accuracy, and has never bothered to read the standard before making
representations about what it guarantees and what it does not. I obtained copies of the ECMA
and ISO/IEC DOCX standards; they are available at
https://www.reginfo.gov/public/do/PRAViewDocument?ref_nbr=202309-0651-002 Since 2019,
the USPTO has contradicted the standards over an over, never citing any evidence for its
opinions. Please take the steps that honest competent people take to be accurate.
1. In 2023, several commenters noted that “Save As DOCX” out of LibreOffice, Google
Docs, Mac Pages, etc. does not result in a DOCX file that mirrors the content of the
native output of the word processor—Greek letters, formulas, paragraph formatting, etc.
change. The USPTO claims “DOCX format is supported by many popular word
processing applications, such as Google Docs, LibreOffice and Pages for Mac.”36 One
has nothing to do with the other. Please explain how this is relevant. “Supported” does
not mean “consistent output.” The issue is reliability of conversion. The Library of
Congress, tested reliability of conversion, and concluded that conversion between wordprocessing formats is not reliable.37 Kindly explain any basis to disagree with the Library
of Congress.
2. In computing a failure rate, the USPTO includes in the denominator “national stage entry
applications and provisional applications” which need not be filed in DOCX, in addition to
the nonprovisional 111(a) applications.38 Adding specious counts to a denominator is
one classic way that dishonest people cheat with statistics. Kindly explain a valid basis
for including “national stage entry applications and provisional applications” in this
denominator and how an honest person could have made this error.
3. In the December 2023 Supporting Statement, the USPTO claims “DOCX… provide[s] a
reliable source of structured text”39 The 2023 Supporting Statement claimed “DOCX
presents documents, including text and formatting, in a manner that is independent of
software, hardware, or operating systems.”40 (a) Kindly identify a statement in ISO/IEC
29500 that corresponds to either claim, with a scope of guarantee that includes DOCX
files generated by non-Microsoft word processors. (b) Kindly reconcile these two claims
with ISO/IEC 29500’s extensive list of “implementation defined” features, particularly
those in the comment letters.41 (c) Kindly reconcile these claims with the ISO/IEC

36

Supporting Statement, note 5 supra, Response to Comment 3.

37

See note 35 supra.

38

Supporting Statement, note 5 supra , Response to Comment 3.

39

Supporting Statement, note 5 supra , Response to Comment 3.

40

Supporting Statement, note 5 supra , Response to Comment 2.

41

Seventy-Three Letter, note 32 supra, at pages 15-17.
DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM

David Boundy

Page 13 of 17

U.S. Patent and Trademark Office
June 26, 2026

29500’s support for extensions.42 (d) Kindly reconcile these two statements with
Microsoft’s documentation43 that shows that Microsoft changes the DOCX format several
times a year.

4. The USPTO claims “The DOCX format, being open source, is available to users of most
every document processing application or software options including Google docs and
Apple products.”44 Kindly explain how this is relevant. “Interoperability” and “open
source” are completely different concepts; neither has anything to do with the other. The
issue is that (a) the DOCX standard does not give the necessary guarantees for
interoperability, and (b) the DOCX standard does say that many dozens of behaviors are
implementation defined,” that is, they may vary implementation-to-implementation, and
the standard permits both extensions and subsetting. Those features guarantee that no
two implementations need be consistent. Even the sentence itself is false: ISO standard
IEC-29500 is not “open source.” This sentence suggests ignorance of technological
vocabulary and willful ignorance of the technological issue that was the subject of the
comment..
5. In 2023, the USPTO claimed “However, the DOCX file is itself a stable carrier for the
textual elements of the document.”45 Kindly explain how this is relevant, and not an
evasion of the issue. patent applications are not only ”textual elements.” Patent
applications include mathematical symbols, tables, mathematical formulae, and text
formatting that communicates meaning (the subscript above). The USPTO choice of
wording suggest that the USPTO concedes that DOCX does not function as a “stable
carrier” for complex documents such as patent applications. Kindly explain.
6. In the December 2023 Supporting Statement, the USPTO claims “PDF submissions,
even rich text formatted PDF submissions, do not provide the same benefits to internal
processes as filing in the DOCX format.”46 Kindly identify statements in either the PDF
standard (ISO 32000) or PDF/A standard (ISO 19005-4) that support this claim.
7. In the December 2023 Supporting Statement, the USPTO concedes “The use of fonts
such as specialized fonts, custom fonts, particularly uncommon fonts, or internal fonts
that have similar labels but use different glyphs from standard font typeface, creates
discrepancies during the document validation and conversion processes.”47 From an
applicant’s point of view, this all just works in PDF. Kindly explain how making this not
work in DOCX is compatible with the statutory obligation, 44 U.S.C. § 3506(c)(3)(E), to
“implement [agency paperwork collections] in ways consistent and compatible, to the
maximum extent practicable, with the existing reporting and recordkeeping practices of
those who are to respond”

42

ISO/IEC 29500 OOXML, note 2 supra, at § 9.1.7, §17.14.25, § 17.17.3, § 18.2.7; see discussion
of “extensions” at pages 8 and 10,

43

See note 4 supra.

44

Seventy-Three Letter, note 32 supra, at pages 15-17.

45

Supporting Statement, note 5 supra Response to Comment 2.

46

Supporting Statement, note 5 supra, Response to Comment 3.

47

December 2023 Supporting Statement, note 5 supra, Response to Comment 3.
DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM

David Boundy

Page 14 of 17

U.S. Patent and Trademark Office
June 26, 2026
8. The USPTO claims “preliminary validation is performed on DOCX documents at the time
of upload. The system immediately detects and supplies the applicant with useful error
and warning messages, allowing for adjustments to patent applications earlier in the
process. This saves time, reduces potential costs to applicants and the USPTO, and
prevents delays in processing”48 Kindly explain technological reasons that these same
benefits cannot be achieved by PDF/A (other than that the USPTO chose to ignore its
own “yearlong study” and to implement them for DOCX instead of PDF/A). The USPTO
claims that “this is not a feature that is available for PDF submissions.”49 Other than the
USPTO choosing to not make it available, kindly explain why it could not be made
available for character-based PDF/A as set forth in ISO 19005.
9. The USPTO's December 2023 Supporting Statement claimed that “Some commenters
noted that a DOCX file created on one computer may not appear the same when viewed
with a different program on a different computer” for several different reasons, and that
users had actually observed the discrepancy. The USPTO's response is “The USPTO
has not experienced the issue raised by commenters.”50 The USPTO claims that “The
USPTO is simply not seeing the unreliability asserted by the commenter.”51 Kindly
explain why personal unawareness is a relevant criterion, to rebut actual observations by
users. Kindly explain what testing and verification methods the USPTO is using, that
would allow it to detect when its rendering of a DOCX file differs from what appeared on
the user’s screen (especially if that user is using a non-Microsoft word processor).
Kindly explain how the USPTO reviews PDF documents that would have been
generated on the user’s computer to serve as a error detection and correction baseline..
Kindly explain how a rare but catastrophic problem is ameliorated by a claim of personal
unawareness.
10. The USPTO claimed that DOCX is suitable because it is a “standard,” yet, comment
letters note, the USPTO refuses to implement the full standard. A partial implementation
cannot claim to be “standard conforming.” The comment letters challenged the
USPTO’s inconsistency. The December 2023 Supporting Statement claims “The
comment letters] suggest that it forces non-standard usage of DOCX files and that the
USPTO is making its own standard.” 52 I am unable to locate any comment that
conforms to either half of this paraphrase. To all appearances, the comment was
misparaphrased to allow the USPTO to evade direct response. Kindly identify any
comment document that made either the “making its own standard” or “forces nonstandard usage.” (Not permitting full exercise of the standard is a completely different
thing than “forces non-standard usage” –perhaps this is another result of the USPTO’s
naïveté in respect to technological terms and standards.) Kindly identify any basis to
reconcile the USPTO’s justification for the rule based on DOCX being a “standard” but
simultaneously refusing to implement that standard. The USPTO admits “equations
presented in certain fonts in DOCX filings may not render accurately in the downstream

48

December 2023 Supporting Statement, note 5 supra, Response to Comment 4.

49

December 2023 Supporting Statement, note 5 supra, Response to Comment 4.

50

December 2023 Supporting Statement, note 5 supra, Response to Comment 7.

51

December 2023 Supporting Statement, note 5 supra, Response to comment 3.

52

December 2023 Supporting Statement, note 5 supra, Comment 8.
DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM

David Boundy

Page 15 of 17

U.S. Patent and Trademark Office
June 26, 2026

documents generated by the USPTO”53 even though those equations are within the
standard. Kindly explain how disallowing or erroneously rendering legal and standardconforming equations can be consistent with the USPTO’s representations that DOCX is
a permissible filing format because it is “standard.”

11. Comment letters noted that the USPTO had relied on a “yearlong study”, which
concluded that PDF is “the right approach,” but (a) had not disclosed that “yearlong
study” during notice and comment, and (b) had acted 180° opposite the conclusion of
the only study cited. The December 2023 Supporting Statement responds as follows:
The USPTO made the decision to encourage submissions of applicant files
via DOCX format in order to leverage the structured text and not rely upon
the OCR process of the PDF conversion. This decision was reached based
on internal IT considerations and initiatives, external sources, and industry
practices, not solely based on any one study.54
Kindly identify anywhere in the NPRM or Final Rule that the USPTO relied on “internal IT
considerations and initiatives, external sources, and industry practices” other than the
“yearlong study.” Kindly identify where the USPTO made available to the public, during
the 2019 notice-and-comment period, any of the “internal IT considerations and
initiatives, external sources, and industry practices” on which the USPTO relied. Kindly
identify any law that grants the USPTO an exemption from the requirement that
agencies disclose the sources they rely on.55
12. In the December 2023 Supporting Statement, the USPTO acknowledges that public
comments estimated that 0.5 hours was too low. The public’s estimates were
accompanied by several pages of explanation. The USPTO responded “the USPTO
considered the most recent feedback received and maintains that the estimated time
burden of 30 minutes is an adequate amount of time”56 with no further evidence or
explanation. The Paperwork Reduction Act requires agencies to support their estimates
with a supporting record, 44 U.S.C. § 3506(c)(3) and 5 C.F.R. § 1320.9, not personal
opinion or “maintains.” Kindly explain the methodology supporting the estimate of 0.5

53

December 2023 Supporting Statement, note 5 supra, Comment 8.

54

December 2023 Supporting Statement, note 5- supra, Comment 11.

55

PTO’s Information Quality Guidelines, § VII(B) (“when asked the USPTO does provide disclosure
of the data sources that have been used and the specific quantitative methods and assumptions (if
any) that have been employed.”); Chamber of Commerce v. Securities & Exchange Comm’n, 443
F.3d 890,899, 901–02 (D.C. Cir. 2006) (“[a]mong the information that must be revealed for public
evaluation are the ‘technical studies and data’ upon which the agency relies [in its rulemaking]”—
agency rule vacated where agency relied on undisclosed extra-record materials in arriving at its cost
estimates); Engine Mfrs’ Ass’n v. EPA, 20 F.3d 1177, 1181–82 (D.C. Cir. 1994) (R.B. Ginsberg, J.)
(APA requires agency to make available “data and studies in intelligible form so that public sees
‘accurate picture of reasoning’ used by agency to develop proposed rule”); Solite Corp. v. EPA, 952
F.2d 473, 484 (D.C. Cir. 1991) (“Integral to the notice requirement is the agency’s duty ‘to identify
and make available technical studies and data that it has employed in reaching the decisions to
propose particular rules… An agency commits serious procedural error when it fails to reveal
portions of the technical basis for a proposed rule in time to allow for meaningful commentary.’”)
56

December 2023 Supporting Statement, note 5 supra, Response to comment 13.
DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM

David Boundy

Page 16 of 17

U.S. Patent and Trademark Office
June 26, 2026
hours to achieve the same reliability as PDF filing. Explain the source of the
USPTO’s expertise to estimate what users do, and how they evaluate error. Explain
with particularity why the USPTO disagrees with the following estimates:
• David Boundy, objectID=107472802 at pages 4-5; estimates total $200 million/yr.
• Brad Forrest, objectID=135639300 at pages 99-107; estimates total $200
million/yr.
• Carl Oppedahl on behalf of 107 co-signatories, objectID=135640900 at page 9;
estimates total $600 million/yr.
• Dr. Cathy Cretsinger, objectID=136718302, at pages 7-8; estimates total $600
million/yr.
13. The public noted that the USPTO’s burden estimates were too low because multiple
categories of burden required by 44 U.S.C. § 3502(2) and 5 C.F.R. § 1320.3(b) were
omitted. The December 2023 Supporting Statement responds that those categories “are
outside the scope of this information collection.”57 Kindly identify a definition of “burden”
that permits exclusion of the relevant categories of burden for “the scope of this
information collection.” For a reminder, 5 C.F.R. § 1320.3(b) required the Office to
account for the following classes of burden:

(i) Reviewing instructions;
(ii) Developing, acquiring, installing, and utilizing technology and systems for the
purpose of collecting, validating, and verifying information;
(iii) Developing, acquiring, installing, and utilizing technology and systems for the
purpose of processing and maintaining information;
(iv) Developing, acquiring, installing, and utilizing technology and systems for the
purpose of disclosing and providing information;
(v) Adjusting the existing ways to comply with any previously applicable
instructions and requirements;
(vi) Training personnel to be able to respond to a collection of information;
(vii) Searching data sources;
(viii) Completing and reviewing the collection of information; and
(ix) Transmitting, or otherwise disclosing the information.
14. The public comments noted that the combination of an auxiliary PDF and DOCX is
unnecessarily duplicative. The USPTO responded “The submission of an auxiliary PDF
is a voluntary action on the part of the applicant. The USPTO does not require
submission of the auxiliary PDF.”58 44 U.S.C. § 3506(c)(1)(B)(iii) and 5 C.F.R.
§ 1320.3(c) require the USPTO to consider burden “whether responses … are
voluntary… or mandatory.” Kindly explain why “voluntary action” is an exemption from
any requirement of the Paperwork Reduction Act, and particularly why “voluntary” is an

57

December 2023 Supporting Statement, note 5 supra, Response to comment 3.

58

December 2023 Supporting Statement, note 5 supra, Response to comment 13.
DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM

David Boundy

Page 17 of 17

U.S. Patent and Trademark Office
June 26, 2026
exemption from the Act’s bar against “unnecessarily duplicative” collection of
information.

David Boundy

June 26, 2026
Date

DAVID BOUNDY | P.O. BOX 590638 | NEWTON MA 02459
T: +1 646.472.9737 WWW .CAMBRIDGETECHLAW .COM